JPH06102347A - Method and device for processing digitally coded pulse signal - Google Patents

Abstract

PURPOSE: To attenuate the effect of a side lobe and eliminate the effect as necessary by comparing the level of filtered received signals matching the transmitting signal of a radar with a detected threshold. CONSTITUTION: A radar receiver is provided with an A/D converter 1 and the output signal of the converter l is distributed to processing channels A and B. The channels A and B respectively have correlators 2a and 2b, filters 3a and 3b, and detecting units 4a and 4b and join together through an AND gate 5. The process A contains auto-correlation and the process B contains auto- correlation with a mismatch code having a length equal to 3N (N: number of instants). The crosscorrelation function between the mismatch code and transmitted codes gives one main central lobe and at least N side lobes which are zero on either one side of the main lobe. The signals from the processes A and B are sent to a level comparator 8 and the output signals of the gate 5 and comparator 8 are sent to an AND gate 9. Only when the output signal of the gate 9 is positive, detection is performed.

Description

Detailed Description of the Invention

[0001]

FIELD OF THE INVENTION This invention relates to the field of radars using digital pulse compression, and more particularly to short range radars.

However, the range is expandable and can be used in analog pulse compression radars and communication receivers that receive low level signals to noise.

[0003]

2. Description of the Prior Art It is known that short range radars emit a series of pulses of width Nτ. Where N is an integer and τ is the elementary instant of the pulse.
Each of the elementary instants of the pulse is modulated with a change in transmission parameter. This parameter is for example phase or frequency. Therefore, for example, the N of one pulse of width Nτ is
Each of the instants may have a phase shift of π. As is well known, phase 0 (coefficient 1) becomes binary code 1,
The position π (coefficient -1) corresponds to the binary number 0.

The received signal, which contains noise and echoes from the target, is compared to the transmitted wave by passing through a correlator.

The correlation of two functions gives a measure of the similarity between them. It is essentially a method of comparison.

Digitally, the signals are represented by a number of sequences that are successively taken samples of one continuous waveform.

A digital sequence f (k) obtained by sampling two continuous functions f (t) and g (t)
And the correlation function of g (k) is expressed by the following equation.

[0008]

[Equation 1] However, if i = sample index k = shift index f and g are different functions, there is cross-correlation.

If f and g are the same function, they are autocorrelated.

The autocorrelation S of the N instant codes represented by the sequence C _{i, where} i is 0 to N-1.
The function of (k) is a measure of the similarity between that code and its k instant shifted ones.

[0011] The result of the autocorrelation of a code with N instant widths is a series of numbers representing the amplitude of the signal as a function of the continuous shifts applied, for which D = Ct / 2 (10m ^{Since << = >>} 66.66 ns), it is possible to scale with respect to time or distance.

The autocorrelation function is symmetric. That is, S (k) = S (-k).

This gives the main peak and side lobes.

Therefore, the code 11101 having five instants and the code 11101 shifted by 0
Since there are 5 similar terms between, S (0) = 5.

To calculate S (1) Since there are two similarities and two differences, S (1) = 0.

To calculate S (2) Since there are two similarities and one difference, S (2) = 1.

Similarly, S (3) = 0 and S (4) = 1.

A number similar to that of a code with N instants shifted in time is the maximum at shift 0 and is equal to N.

S (0) = N S (0) is the main peak and its width is equal to one shift. This gives the target distance.

For other shifts where k is in the range 1 to N-1, the autocorrelation gives a small result which can be zero.
This result depends only on the chosen code. The result of them is a side lobe. They are symmetrical on either side of the main peak. S (k) is equal to S (-k).

If the sidelobe of a code with N instants has the value m, its attenuation (dB) with respect to the main lobe is equal to lobe i = 20log m / N.

For an example code with five instants 11101, the time representation of its autocorrelation function is shown in FIG. 1 with k from -5 to +5.

This function shows the main lobe for k = 0 and the four side lobes of value 1.

The relative level of the side lobe is 20 log (1
/ 5) = 14 dB.

An example of a two-phase code is shown.

The two-phase code is only a special case of the polyphase code when the coefficient is a complex number (nth square root of 1).

Two-Phase Codes: 1 -1 4 Phase Codes: 1 j -1 -j There are other ways of limiting the correlation when using complex planes.

The phase code with N instant lengths used in the transmission can be represented in a complex plane in the form of an N-1 polynomial C (z) limited by the variable z ^-1 as is well known. This variable represents one shift in time when the rank of the sample and the phase of the code define the coefficient (+1 or -1 for a two-phase code).

2 with 5 instants 11101
The phase code can be represented in the form of a fourth order polynomial.

C (z) = 1 + z ^-1 + z ^-2 -z ^-3 + z ^{-4 The} reference code CR (z) applied to the code C (z) is the conjugate complex of C (z) inverted in time. Corresponds to the code. In fact, when the code C (z) is autocorrelated with its replica CR (z) in the digital correlator, this replica is the conjugate of the initial code, which has been inverted in time. That is, C (z) = 1 + z ^-1 + z ^-2 -z ^-3 + z ^-4 .

Code 11101 first receives 1 (stage 1 of the digital correlator), then -1 (stage 2) and finally 3x1 (stages 3, 4, 5). Therefore, CR (z) = 1-z ^-1 + z ^-2 + z ^-3 + z ^{-4 This} physically ^explainable operation is equivalent to conversion of z to 1: Z and addition of delay time. In the case of a polyphase code, it is necessary to use the following conjugate coefficient.

CR (z) = C ^* (1 / z) z- ^(n-1) That is, the code is 11 j-j-1.

This can be expressed as follows.

C (z) = 1 + z ⁻¹ + jz ⁻² −jz ⁻³ −z ⁻⁴ This matched code is CR (z) = − 1 + jz ⁻¹ −jz.
^{It is -2} + z ^-3 + z ^-4 .

In the correlator, these coefficients are arranged as follows.

The stage 1 2 3 4 5 coefficient −1 j −j 1 1 1 correlation function S (z) is expressed as follows.

S (z) = C (z) .CR (z) = C (z) .C ^* (1/2) .z- ^{(n-1) In} the example of the code 11101, S (z) = C ( z) · CR (z) = (1 + z ⁻¹ + z ⁻² −z ⁻³ + z ⁻⁴ ) (1-z ⁻¹ + z ⁻² + z ⁻³ + z ⁻⁴ ) S (z) = 1 + z ⁻² + 5z ⁻⁴ + Z ^-6 + z ^{-8 The} result obtained in this example, where the same code is merely two-phase, is again 101050101, the main peak is 5 and the sidelobe is 1.

The correlation operation is performed for every distance in the radar range, ie every Cτ / 2. This means that the correlation operation should be performed every τ. The corresponding polynomial therefore uses a factor z ^−m which represents a different delay.

Radar designers use pulse compression to improve resolution over distance.

FIG. 2B shows the response obtained at the output of the matched filter, starting with the standard radar that emits pulses of width T shown in FIG. 2A.

In these figures, the square wave signal of width T corresponds to a sawtooth autocorrelation function which introduces uncertainty in the distance of the target.

An ideal response can be obtained by making a pulse compression radar that emits pulses of width T = Nτ.

The autocorrelation function corresponding to the transmitted signal is shown in FIG.
Shown in (a) and FIG. 3 (b).

The transmitted signal is five instants 11101.
, Whose autocorrelation function is represented by a peak of width τ. The uncertainty over distance is N times lower.

As described above, this ideal response is obtained, but the side lobes are different.

The detection result is compared with a threshold value after a process determined by the radar type for each distance, and all values larger than this threshold value are detected. Therefore,
Under some conditions (very large echo) side lobes can be detected. This unwanted detection is echo stimulated but at the wrong distance.

The indication of the target is activated by this detection. Therefore, this detection problem is related to the display. The measurements and calculations made in this process localize the target both in distance and direction, and in Doppler radar allow them to accurately ascertain the velocity of the detected target. These information elements can be displayed on a CRT or digital screen.

The detected target pip generally represents the width of the transmitted pulse for standard radar and the width of the prime instant for pulse compression radar, respectively.

Therefore, in the case of pulse compression (CI),
The side lobes also cause a pip on the screen from some signal portion reflected by the target. This phenomenon occurs as soon as the S / N ratio of the echo exceeds the sidelobe rejection.

The side lobe is 2N with the width of a plain instant
Distributed over a distance equal to double, centered on the main lobe. Their number therefore increases with increasing compression ratio.

From this, it can be seen that autocorrelation inevitably causes side lobes. Their level depends on the code used, so the choice of code is important.

For a code with N instants, the main lobe always has the value N.

It has been considered to minimize the sidelobe value. The minimum value that can be obtained is 1. A code that satisfies the condition that the maximum number of side lobes is 1 is called a Barker code.

A study of this code shows that they only exist up to a maximum length of 13, and that the level of sidelobes for that value is therefore 20log 13 = 2.
It was found to be 2 dB.

It is also conceivable to increase the value of the main peak. To achieve this, N should be increased. Pseudo-random sequences or longest sequences are particularly important because they have a structure similar to that of random sequences and have useful autocorrelation functions. Loop shift registers simply and practically generate pseudo-random sequences. After the stage of the shift register is initialized with 1, the add modulus 2 is made at the output, and the result of this add constitutes the input of that register. The output of this shift register constitutes a generator of a repeated 0 and 1 sequence. These sequences have a length determined by the loop. When the summed stages are chosen appropriately, a sequence with a maximum length of L = 2 ^N -1 is obtained for N stage registers.

The result of the autocorrelation function of these codes is N
^It gives the ratio of the main peak to side lobes of about ^1/2 .

In practice, such a code can reduce the side lobes below the main peak by less than 24 dB.

This method results in losses that have to be calculated in each case. Very satisfactory 0.2
Results are obtained with losses on the order of ~ 1 dB.

All of the above methods used to reduce side lobes have the following characteristic quality deficiencies.

A) With a Barker code (length N), 2
A relatively high level of 2 distributed over N instants
The next lobe remains.

B) In the mismatch code, secondary lobes that are distributed at a low level but in a larger range remain.

The resulting attenuation is generally sufficient for long range radar. Short-range radars, on the other hand, generally have a very large signal dynamic range.

Undesirable detection by secondary lobes, which is especially noticeable in short range radars, has many drawbacks, especially when making radars with special functions such as, for example, acquisition and tracking of moving targets. When the target is determined, its position is determined and the radar continues to point at it.
It moves in both distance and relative direction and remains locked on as long as the operator wants it.

[0064]

This method works correctly only if the target is correctly localized. The presence of side lobes makes tracking moving targets difficult. Similarly, there are zones (eg roads) where the presence of moving targets is normal when the protected site is to be monitored by radar. In this case, no alarm should be raised by detection, and these zones should not be barred. Boundary line monitoring is one example of zone management. Radar monitors a portion of the ground for a predetermined width (eg, a protected site edge) and alarms only for targets moving within this zone. Targets in the forbidden zone facilitate detection in adjacent zones by their side lobes in the active monitor zone, and promote unwanted detection.

It is therefore an object of the present invention to provide a method of attenuating the effects of side lobes and eliminating them if necessary.

[0066]

Thus, the present invention is a method of processing a signal received directly or indirectly from a transmitter which emits pulses digitally encoded according to a transmission code of N instants. This processing method comprises a filtering step matched to the transmitted signal and a detection step by comparison of the filtered level of the signal with a so-called detection threshold, the received signal being subjected to two processes A and B,
Process A includes autocorrelation and process B includes cross-correlation with a mismatch code of length equal to at least 3N, the mismatch code and transmit code cross-correlation function being one of the main central lobes and this main lobe. Give at least N sidelobes that are zero on either side. It is determined that there is a detection associated with one of the correlation or cross-correlation instants only when the functional level of this same instant of autocorrelation and cross-correlation is above the detection threshold.

[0067]

This signal, even when propagating between the transmitter and the receiver in the corridor by reflection or refraction in layers of the atmosphere, enters directly from the transmitter in the case of a signal transmitted for communication purposes. It is believed that

This signal is received after the transmitter and the receiver are in close proximity to each other and the receiver reflects the signal sent from the transmitter at a non-cooperative target, which is the case for radar signals. It is considered to be indirect. The instants of the autocorrelation and cross-correlation functions are the same if they are at the same distance from their main lobes.

In addition to this measure of detection, it is determined that there will be a detection at the end of each of processes A and B only after the signals from processes A and B have been processed for an additional measure. This measure is the level difference between the two signals that is less than some threshold.

The transmitter operating according to the method of the invention has two channels for the processing of the received signal at the filtering level of the signal, one of which contains the autocorrelation and the other of which the mismatch length is at least equal to 3N. It includes a cross-correlation with the signal, the cross-correlation function having at least N zero sidelobes on either side of the main lobe. The detection circuit also has two channels, each including a detection threshold, a comparator for the signal and a logic module for issuing a signal corresponding to the detection only when there is a detection in both comparators at each instant. Given an additional measure of this signal, the chain of processing includes logic modules or circuits that can make the desired comparisons.

These two processing channels, the filtering level and the detection level, can be physical or temporal. If processes A and B are performed successively in multiple modes on one physical channel, they are temporal.

[0072]

The principle of the present invention can be expressed in the following three points. (1) Sending and receiving N instant codes. Two
One processing channel.

^* Standard Processing "A" The N coefficients of the correlator are the same as those of the code transmitted. After filtering, the detection information element PIPA _j is available for each instant j.

PIPA _j = 0 When no instant is detected.

When there is PIPA _j = 1 detection.

There can be detection of only N instants on either side of the main peak. (2) Weighting process "B" using a mismatch code having a length equal to at least 3N upon reception. In this case, the coefficients are calculated to obtain at least N 0's (zero quadratic lobes) on either side of the main peak. The detection information elements PIPB _j can be used after filtering as in process "A".

PIPB _j = 0 When there is no detection.

PIPB _j = 1 In case of detection.

Detection is possible only in the main peak and in the two zones located N instants before or after this peak. (3) The resulting pip PIP _j is obtained by ANDing the detected information elements PIPA _j and PIPB _j produced by the two processes A and B.

PIP _J = PIPA _J “AND” PIPB _J The amplitude of the signal for each of the instants after the AND operation in the correlation and cross-correlation outputs is shown in FIG.
Shown in (b) and (c), respectively.

Due to the nature of the chosen code, there is detection only for the main lobe.

In the case of radar, the instants representing physically consecutive clock increments correspond to range windows. On the other hand, in communication they correspond to a time shift in the receiving operation of the signal to be received.

In the case of the Doppler radar, the fast Fourier transform (FFT) in each of the processing channels A and B is the speed i and the distance j of the detection information elements PIPAi _j and PIPBi _{j which} the AND circuit converts into PIP _j detection information.
Allows a decision about each window in.

The calculation mode of the mismatch code will be described in detail below.

The transmitted pulse is 2 by N instants.
It is a pulse formed with a phase modulated at the rate of a binary code. The transmit code is chosen to give the best sidelobe rejection. If the length N is less than 13, the Barker code is the best.

Standard processing A that satisfies the match code at the time of reception
Is used. The coefficients of the correlators are therefore identical except for the sign. The result obtained is a maximum of 22 dB of sidelobe rejection for the above, ie, 13 instant Barker codes. These side lobes exit at N instants on either side of the main peak.

The minimum length of the mismatch code CD (z) used in process B and required to create N zero sidelobes on either side of the main peak is 3N.
This code is obtained in two steps.

(1) Calculation of mismatch code.

(2) Additional calculation of the coefficients to obtain zero for each of the N lobes surrounding the main peak.

The purpose of the mismatch code CD (z) is to obtain an ideal correlation function without sidelobes.

S (z) = N · z−m Therefore, we only have to find the code CD (z) as follows.

S (z) = C (z) · CD (z) = N · z ^−m However, CD (z) = N · z ^−m · 1 / C (z) CD (z) is z ^−m only Corresponds to the inverse filter 1 / C (z) delayed and amplified by a factor N. Furthermore,
Similar to the match code, we need to invert C (z) over time to have the coefficients in the right array.

Z _i (i = 1 to N-1) is a polynomial C (z)
Assuming that there are N-1 complex roots of CD (z), the delay z ^−m added or subtracted can be written in the following form.

CD (z) = 1 / (z ⁻¹ −a ₁ ) · (z ⁻¹ −a ₂ ) ... (z ⁻¹ −a _n ⁻¹ ) C (z) has two groups of roots, namely It is divided into C ₁ (z) for roots within a unit circle and C ₂ (z) for those with modules greater than one.

CD (z) = 1 / C ₁ (z) · 1 / C ₂ (z) For C ₁ (z), 1 / (z ⁻¹ −
The approximation of a) can be found.

1 / (z ⁻¹ −a) = (a ⁻¹ + z ⁻¹ · a ⁻² + z ⁻² · a ⁻³ + ... z ⁻ⁱ · a ^{−a (i + 1)} ) This sequence is a module If a is smaller than 1, it converges. Therefore, this can be applied to the roots of the first group of C (z). This gives the polynomials that make up the approximation. Its accuracy increases with i.

For roots with modules greater than 1, this sequence diverges. Variable conversion is performed to converge this.

Y = 1 / z ⁻¹ = z ⁻¹ = 1 / y 1 / (z ⁻¹ −a) = − y / a · (1 / y−a ⁻¹ ) The second term is a ^−1. Since the module is less than 1, it can be divided into convergence sequences. If y is replaced by its value z, the following approximate expression is obtained.

1 / (z ⁻¹ −a) = z + az ² + a ² z ³ + ... + a ⁱ z ^{i + 1} This method depends on the left division. Other polynomials corresponding to module roots greater than one can then be determined.

The product of these N-1 polynomials in limiting the number of terms to the desired length gives the desired mismatch code equation.

The coefficient of the mismatch code is the result of several approximations, and the final result depends on the calculation accuracy. Furthermore, the coefficients found should be coded for the number of bits determined by the chosen correlator.

The result of cross-correlation with a mismatch of one code gives a low but non-zero sidelobe. To remove them, we need to repeat the calculation for each instant of the code starting from the main peak as follows. First, the center N coefficients corresponding to the main peak are retained. Then, after shifting, the value of the next coefficient is calculated to obtain a zero correlation result, and this operation is continued during N shifts to obtain N subsequent coefficients. It can be inferred from the latter because the N coefficients before the main peak have the same module and the same sign as the corresponding coefficient of the mismatch code.

The code thus obtained is composed of at least 3N coefficients.

Code 11: An example of calculation of coefficient weighting
101 will be described next.

It is assumed that this code is used to code the transmitted pulse and perform autocorrelation according to process A.

The coefficient CD of the mismatch code encoded by 8 bits has the following decimal value.

C ₀ C ₁ C ₂ C ₃ C ₄ C ₅ C ₆ C ₇ -3-11 29 -31 -5 74 -127 79 C ₈ C ₉ C ₁₀ C ₁₁ C ₁₂ C ₁₃ C ₁₄ 127 745- 31-29-11 3 Coefficient CB that gives five zeros on either side of the main peak
Is inferred from the CD based on the 5 central coefficients. With an input signal standardized with an amplitude of 1, the correlator calculation for one shift corresponds to

Coefficient C ₅ C ₆ C ₇ C ₈ C ₉ C ₁₀ 74 -127 79 127 127 74 X Signal The output signal of this correlator is as follows.

S (1) = C ₆ -C ₇ + C ₈ + C ₉ + C ₁₀ = 0 C ₁₀ = -C ₆ + C ₇ -C ₈ -C ₉ = 127 + 79-127-74 = 5 Therefore, in this case, The coefficient corrected in the calculation is equal to the code originally found for the mismatch code CD. This is not always the case.

Performing this calculation for 2, 3, 4, and 5 shifts determines the following coefficients:

A portion of a radar receiver operating in accordance with the present invention is shown in FIG.

This part has an analog / digital converter (A / D converter) 1.

The output signal from this converter 1 is then split between channel A and channel B. Each of these two channels has a correlator 2a, 2b, if necessary a filter 3a, 3b, and a detection unit 4a, 4b, which are merged by an AND gate 5 shown in FIG.

This embodiment of the radar according to the invention relates to a Doppler radar for digital pulse compression, where each of the processing channels is in addition to a correlator and a filter, as shown in FIG. 6, FFT modules 6a, 6b and a post. Includes integration modules 7a, 7b.

FIG. 7 shows the output section of the processing channels A and B as shown in FIG. 5 or 6. This figure shows a module 8 allowing the introduction of other measures. In the example of FIG. 7, this module is a level comparator. The signals from each of processing channels A and B are therefore initially AN
It is sent to D-gate 5 and then to module 8. AN
The output signals of the D gate 5 and the module 8 are AND gates 9.
Sent to. The detection is performed only when the output signal of the AND gate 9 is positive.

Although these modules are represented by the physical units of FIG. 7, they can be functionally integrated into a logic module.

As described above, the signal processing method according to the present invention can be extended to a communication receiver by using the apparatus shown in FIG. 5, for example.

An exemplary embodiment is a 3N length mismatch code, which is the minimum to obtain the cross-correlation function of the transmitted code containing N zero side lobes on either side of the main lobe. Is the length. Therefore, a mismatch lobe with a length exceeding 3N can be taken.

However, this length is mainly controlled by the elements that are actually on the market, especially the special processors used for shift registers, and then the increase in processing time.

[Brief description of drawings]

FIG. 1 is a diagram showing an autocorrelation function.

FIG. 2 is a diagram showing standard radar transmission pulses.

FIG. 3 is a diagram showing an autocorrelation function for a transmission signal.

4A shows an output of a process A, and FIG. 4B shows a process B.
FIG. 3C is a diagram showing the output of FIG. 3C, and FIG.

FIG. 5 is a block diagram showing a portion of a receiver according to the present invention.

FIG. 6 is a block diagram showing a portion of a Doppler radar receiver according to the present invention.

FIG. 7 is a block diagram showing the output ends of processing channels A and B.

[Explanation of symbols]

 1 analog / digital converter 2a, 2b correlator 3a, 3b filter 4a, 4b detection unit 5, 9 AND gate 6a, 6b FFT module 7a, 7b post-integration module 8 level comparator

Claims (7)

[Claims] 1. A method of processing a signal received, directly or indirectly, from a transmitter that sends digitally encoded output pulses according to N instant transmission codes, the filtering step matching the transmitted signal. , And a detection step by comparison of the filtered level of said signal with a so-called detection threshold, said received signal being processed by two processes A and B, said process A performing an autocorrelation process. Including, said process B comprises a cross-correlation process with a mismatch code of a length equal to at least 3N, the cross-correlation function of this mismatch code and the transmitted code being one main central lobe and either side of this main lobe. Given at least N sidelobes that are zero, and when the function level of this same instant of autocorrelation and crosscorrelation is above the detection threshold Only determined as there is a detection relating to one of the correlation or cross-correlation instant processing method of digitally encoded pulse signal. 2. The method of claim 1, wherein it is determined that there is a detection only after the signals from said processes A and B are compared with an additional measure. 3. The method of claim 2, wherein the additional measure is a level difference between the two signals that is less than a certain threshold. 4. A signal receiver for processing a received signal, comprising at least one correlator and one detector in series, each of which comprises two channels A and B which merge in an AND circuit. 5. The signal applied to the channels A and B is from an analog / digital converter.
Receiver. 6. Each of said channels A and B further comprises an FFT.
The Doppler radar receiver of claim 5 including a module and a post-integration module. 7. The signal from each of said processing channels A and B is fed to a comparison module, said A receiving said output signal of said comparison module and the signals from channels A and B.
7. The radar receiver according to claim 4, wherein the signal of the ND gate is given to another AND gate.